Regression-Based Tests of the Market Pricing of Accounting Numbers: The Mishkin Test and Ordinary Least Squares

Authors


  • We have benefited from comments and suggestions made by seminar participants at The University of North Carolina and the University of Rochester, and by Charles Plosser, Bill Schwert, Ross Watts, Chuck Whiteman, an anonymous referee, and Ray Ball (editor). We gratefully acknowledge the financial support of the London Business School, Penn State University, and the Simon School of Business at the University of Rochester. This paper was previously circulated under the title “On the Use of Mishkin's Rational Expectations Approach to Testing Efficient-Markets Hypotheses in Accounting Research.”

ABSTRACT

The test developed in Mishkin [1983] (hereafter, MT) is widely used to test the rational pricing of accounting numbers. However, contrary to the perception in the accounting literature, the exclusion of variables from the MT's forecasting and pricing equations leads to an omitted variables problem that affects inferences about the rational pricing of accounting variables. Only if the omitted variables are rationally priced is their exclusion irrelevant. Failure to recognize this issue leads accounting researchers to employ the MT without appreciating how omitted variables affect the inferences they draw. We demonstrate that when additional explanatory variables are included in the MT, the rational pricing of accruals is not rejected. That is, the accrual anomaly documented in Sloan [1996] vanishes when additional explanatory variables are incorporated into the MT. We also show that in accounting research settings, where samples are large, ordinary least squares (OLS) is equivalent to the MT. As a result, accounting researchers should consider using OLS or be more explicit about the exact advantages of the MT over OLS in their research setting.

Ancillary